F1 Test Drive

This runs end-to-end validation of the Cyrus agent system by spinning up an F1 server, creating test issues, starting agent sessions, and verifying that activities flow correctly from issue tracker through EdgeWorker to rendering. You'd reach for this before merging changes to core runner or harness code to catch integration breaks early. The protocol is methodical: setup a test repo, verify each component in sequence, optionally test the Slack chat path with synthetic events, then write up results in a structured markdown report. The pass/fail criteria are practical and the skill explicitly stays harness agnostic so both Claude subagents and other workflows can reference the same test protocol. It's manual QA codified into repeatable steps with clear exit conditions.

Install to Claude Code

npx -y skills add ceedaragents/cyrus --skill f1-test-drive --agent claude-code